JP3051914B2 - Unequal error protection method based on multi-block coded modulation - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital modulation / demodulation technique for wireless / wired transmission using a digital signal, and more specifically, to provide non-uniform error protection by paying attention to the difference in importance of transmitted data bits. The present invention relates to a modulation / demodulation technique for digital signal transmission that enhances transmission quality by performing such a method.

[0002]

2. Description of the Related Art Conventionally, transmission is performed by giving an error rate characteristic equal to data bits transmitted in digital signal transmission. In this case, if the signal-to-noise ratio that can be provided by the transmission line falls below a required value due to any failure in the transmission line, a large error rate that cannot be tolerated in the transmitted data bits occurs, and the transmission quality is uniformly increased. There was a problem that it deteriorated.

[0003] On the other hand, transmitted data bits have different importance. For example, in ATM transmission, the data of the head cell is more important than the data of other cells. By assigning better error rate characteristics to data bits with high importance, transmission quality can be prevented from being uniformly and largely degraded, and overall transmission quality can be improved.
Therefore, it has been necessary to develop a method that can provide such uneven error protection (hereinafter abbreviated as UEP) characteristics.

As a method of generating UEP, there is a method of simply changing the conventional signal point constellation and increasing the MSED of other bits by sacrificing the MSED of some bits. However, this method is limited to the MSED of the original signal point constellation, and there is a problem that the MSED of the data bit of high importance cannot be sufficiently increased or the MSED of the data bit of low importance becomes too small. Was.

[0005] A method of generating a UEP by the trellis coded modulation method can alleviate the above problem. As the simplest method, different trellis coding is performed on data bits having different degrees of importance, and decoding corresponding to each is performed on the decoding side. Obviously, however, this method has a problem that the structure becomes complicated or redundant.

[0006]

Apart from the above, there is also a method of generating UEP using one type of trellis coding / decoding. However, in the methods based on these trellis coded modulation schemes, it is necessary to hold the trellis for a considerably long period of time in order to reduce possible performance degradation due to trellis truncation operation on the decoding side. And the delay due to the processing time must be increased.

[0007]

SUMMARY OF THE INVENTION The present invention has been proposed in view of the above. In a modulation / demodulation system for digital signal transmission, multiple block coded modulation is used, and the MSED of each code level of the multiple block coded modulation is different. UEP is created in one encoding / decoding process by utilizing what is obtained from the approach.

The present invention is also characterized in that Viterbi decoding based on a trellis diagram is realized for a multi-block coded modulation method used to generate UEP.

The present invention further provides, as a technique, raising the MSED of all transmitted data bits to a certain value by multiplex-coding modulation, and without degrading this value. It is characterized in that the MSED can be further increased for some of the bits.

The present invention is further characterized in that, as another method, the UEP based on the multi-block coded modulation is created by using the conventional constellation change.

[0011]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 shows a bit matrix of an encoder according to a first embodiment using an 8-PSK signal point arrangement according to the present invention and a method of generating an 8-PSK signal. However, this assumes that k symbols are assigned to one branch of the trellis. In the figure, the horizontal row is called a code level, and is labeled as l ₁ , l ₂ ,..., L _{k + 2} in order from the top. One vertical column corresponds to one branch of a decoding trellis diagram described later. Therefore, it is necessary to form k symbols from one vertical column. As shown at the bottom of the figure, the k symbols have code levels l ₁ and l 1
Used on their first two bits as a common bit 2 bits from _2, and each of the third bit l _3, l _4,
.., Use bits from l _{k + 2} .

In FIG. 1, a ₁ , a ₂ ,..., A _{6k + 7} are data bits, and c ₁ , c ₂ ,..., _{Ck + 1} are code levels l ₃ , l ₄ ,. _This is a parity check bit at _{k + 2} . The check bits follow the following equation.

[0013]

(Equation 1)

The MSED at each code level is the minimum Hamming distance of each level (hereinafter abbreviated as MHD) and the MS of the subset corresponding to the level when the set is divided.
Obtained from the product of ED. Here, the code levels l ₁ and l ₂
Are used redundantly in k symbols, the total MHD of the level is k k of the MHD of the code of the level.
Double. Therefore, the MSED for each code level can be calculated as:

[0015]

(Equation 2)

Therefore, if k> 2, the code level l ₃
The MSEDs of the code levels l ₁ and l ₂ can be further increased while the MSEDs of the code levels below and are maintained at large values. The ratio (I / A) of the bits included in the code levels l ₁ and l ₂ , that is, the bits whose MSED is large, to the total bits is calculated from the following equation.

[0017]

(Equation 3)

The number of data bits carried per symbol, that is, the throughput (THR) is calculated from the following equation.

[0019]

(Equation 4)

Another major feature of the first embodiment is as follows.
This is applicable to all code configurations irrespective of the value of the number k of symbols assigned to the branch . For example, the number of states
Is 8-PSK as an example . It is shown that an 8-state trellis diagram enables Viterbi decoding using an 8-state trellis diagram.
This will be described with reference to FIG. 2, denoted by S ₁ to S ₈
Each node is a trellis diagram, and the vertical nodes
The maximum number is called the number of states of the demodulation trellis diagram. A _{1 to} A ₄
And B _{1 to} B ₄ are branch variables whose arguments are 8
-8 signal points of PSK (0, 1, ..., 7). The state S ₁ to S ₈ of eight takes a branch variables shown in the lower part, respectively, of FIG.

On the other hand, the number of parallel paths (PARA) included in each branch of the 8-state trellis diagram of FIG. 2 is PARA = 2 ^{k -1} according to the value of ^k . Table 1 shows MSED values at levels l ₁ and l ₂ of each code configuration from k = 1 to 6,
The I / A value, THR value and PARA value are summarized.

When k symbols are assigned to one branch, there are 2 ^{k + 2} combinations of k symbols in total. Derivation of a branch variable follows the following procedure. First, 2-bit values (00, 01, 1) commonly used in each symbol from levels l ₁ and l ₂ are used.
According to (0, 11), the 2 ^{k + 2} combinations are divided into 4 groups of 2 ^k (A ₁ A ₂ , A ₃ A ₄ , B ₁ B ₂ , B ₃ B ₄ ). Next, for each of the above groups, one set (A ₁ , A ₃ ,
B ₁ , B ₃ ), and the set of 1 is set as a set (A ₂ , B ₃ ).
_{A 4, B 2, B 4} ). As described above, eight sets each having 2 ^k−1 combinations are obtained, which are used as branch variables. As an example, a branch variable when k = 4 is shown below.

[0023]

(Equation 5)

FIG. 3 shows an example of the level l ₂ of the first embodiment, where k = 2 obtained by computer simulation.
The BER characteristics at 6 were shown. BE as k increases
It can be seen that R is going to be improved. As k increases, the amount of improvement in BER tends to decrease, which is considered to be due to an increase in the PARA value.

The following table shows the calculated values of the parameters depending on k in the first embodiment of the present invention.

[0026]

[Table 1]

FIG. 4 shows the BER characteristics of each level when k = 6 in the first embodiment. For example, levels l ₁ and l ₂
, A bit of high importance is placed on the other level, and a bit of low importance is placed on the other level. In order to obtain the same BER = 10 ^-4 , the bit of high importance has a small E _b / of about 3.4 dB or more. N ₀ is enough. Where E _b / N ₀ is the signal-to-noise ratio per data bit.

The method of obtaining the UEP by the multi-block coded modulation method has been described above.
A method for obtaining UEP by combining the multi-block coded modulation scheme and changing the signal point constellation will be described. Here, a multi-block coded modulation scheme in which two symbols are allocated to one branch using 16-PSK is used as the second embodiment. FIG. 5 shows a bit matrix and a symbol generation method of the encoder of this example.

M at each code level in the code configuration of FIG.
The SED can be calculated in the same way as described above and follows the equation:

[0030]

(Equation 6)

From the above results, the levels l ₁ and l ₂ have smaller MSEDs than the other levels. However, since the BER of the lower level is limited to the upper level, it is necessary to increase the distance between signal points restricted by the upper level in order to obtain BER to be acquired by the lower level MSED. In order to realize this, the conventional signal point arrangement is changed to a non-uniform signal point arrangement as shown in FIG. By reducing the angle α in FIG. 6, the following four signal point groups G ₁ = {0, 1, ₂ , 3}, G ₂ = {4,
5, 6, 7}, G ₃ = {8, 9, 10, 11} and G
₄ = levels l ₁ and l between {12, 13, 14, 15}
The MSED in ₂ (hereinafter abbreviated as MSED-1) increases, and by setting MSED-1 to the same value as the lower-level MSED value, the BER to be acquired by the lower-level MSED is obtained. On the other hand, by this signal point operation, the levels l ₁ and l ₂ in each of the groups G _{1 to} G ₄ are set.
, The MSED (hereinafter, abbreviated as MSED-2) at the level becomes smaller, and as a result, the bit B at the levels l ₁ and l ₂
ER deteriorates.

The MSED in the signal set corresponding to the code level equal to or lower than the level l ₃ irrespective of the value of the angle α in FIG.
Does not change. On the other hand, the MSED of the signal set corresponding to the levels l ₁ and l ₂ largely depends on the angle α. Level l
₁ and above MSED-1 and the l ₂ MSED-
2 is calculated by the following equation.

[0033]

(Equation 7)

The signal point arrangement in FIG. 6 is the conventional 16-PSK when the angle α is π / 8, but when α is π / 12, the minimum MSED-1 at levels l ₁ and l ₂ is the lower level MSED Has the same value as Table 2 shows that α in Embodiment 2 is π /
The calculated values of MSED-1 and MSED-2 at levels l ₁ and l ₂ at 12 are shown. From the table, both MSE
It can be seen that in order to obtain the minimum value of D-1 of 4.0, the value of MSED-2 decreases from 2.3 in Equation 6 to 1.1.

[0035]

[Table 2]

FIG. 7 shows α = π / 8 and π / 12 of the second embodiment.
An example of the result of computer simulation of the BER of each code level at the time of is shown. When α = π / 8, each code level indicates a close BER, while when α = π / 12, the BERs of the other code levels are greatly different from the code levels l ₁ and l ₂ . When BER = 10 ⁻⁴ , the difference between E _b / N ₀ is about 4.1 dB.

The MSE according to the second embodiment of the present invention
Angles of D-1 and MSED-2 α = π / 8 and π /
The calculated value at the time of 12 is as shown in Table 2 above.

Another feature of the UEP creation method of the second embodiment is that the BER of a large number of data bits is increased by sacrificing the BER of a small number of data bits. In the above example, the MSED is about 79% of the total bits.

The two methods according to the present invention have been described based on the first and second embodiments. However, the present invention is not limited to the above-described embodiments, and has the structure described in the claims. It can be implemented in any way without change. For example, in the first embodiment and the second embodiment,
Although MPSK modulation signals such as 8-PSK and 16-PSK have been mentioned, the non-uniform error protection scheme based on multi-block coded modulation of the present invention works effectively for many modulation schemes such as digital multi-level QAM. .

[0040]

According to the unequal error protection method based on the multi-block coded modulation of the present invention, it is possible to more easily improve the B value for the bits of high importance in the transmitted data bits.
ER characteristics can be provided, and the overall transmission quality can be expected to be improved. Also, by utilizing the fact that a decoded trellis diagram has a block structure and a trellis structure simultaneously, a code is realized while implementing Viterbi decoding. Excellent effects are obtained, such as the length can be made shorter, the memory required for decoding and the number of calculations can be reduced.

[Brief description of the drawings]

FIG. 1 is a conceptual diagram showing a matrix and a symbol generation method of an encoder of a 1-branch k-symbol multi-block coded modulation scheme according to a first embodiment of the present invention.

FIG. 2 is a decoding trellis diagram according to the first embodiment of the present invention.

FIG. 3 is a characteristic diagram showing a bit error rate characteristic at a code level l _{2 in} the first embodiment of the present invention.

FIG. 4 is a characteristic diagram showing a non-uniform error rate characteristic according to the first embodiment of the present invention.

FIG. 5 is a conceptual diagram showing a matrix and symbol generation method of an encoder of a 1-branch 2-symbol multiplex block coded modulation scheme using 16-PSK.

FIG. 6 shows a non-uniform 16-PS according to the second embodiment of the present invention.
It is a K signal point arrangement diagram.

FIG. 7 is a characteristic diagram showing a non-uniform error rate characteristic according to the second embodiment of the present invention.

Continuation of the front page (58) Fields investigated (Int. Cl. ⁷ , DB name) H04L 1/00 H03M 13/23 H04L 27/00 H04L 27/22 H04L 27/38

Claims (4)

(57) [Claims] 1. In a wireless or wired digital signal transmission, since data bits to be transmitted have different degrees of importance, a multi-block coded modulation scheme capable of setting a bit error rate according to the degree of importance of the data bits. Number of symbols used and assigned to one branch of decoding trellis
k (where k> 2), the encoder matrix
Least squares Euclidean distance at each code level
(Hereinafter abbreviated as MSED) is used for multi-block coded modulation.
Bit error rate characteristics at each code level (hereinafter referred to as BER)
Abbreviated) to make the digital signal data bit
Unequal Error Protection <br/> how based on multi-block coded modulation, characterized in that it has to produce the error protection in response to Tsu and. 2. The decoding trellis is assigned to one branch.
By increasing the number of symbols k
The number of symbols k assigned to one branch of the matrix
To increase some of the code in the encoder matrix.
While increasing the signal level MSED, other code levels
Unequal Error Protection how based on multi-block coded modulation according to claim 1, characterized in that so as not to impair the MSED of. 3. A signal point arrangement of a modulation method is changed,
Reduce the MSED of some signal levels in the detector matrix
To increase the MSED of other code levels
Unequal Error Protection side based on multi-block coded modulation according to the to claim 1 or claim 2, wherein as
Law . 4. When 8-PSK signal point arrangement is used.
The number of multiplexed block coded modulations (divided into one branch)
Demodulated trellis line regardless of the number of assigned symbols k)
Unequal Error Protection how based on multi-block coded modulation according to any one of claims 1 to 3, wherein the number of states in FIG. Is constant.